test: harden reconnect-restart-in-place test against dash-spv lock-release race

[Claudius-Maginificent]

Why this PR exists

• Problem: context::wallet_lifecycle::tests::reconnect_restart_in_place_reuses_backend intermittently fails CI with SpvError("Data directory locked: ... already in use by another process"). Confirmed pre-existing on v1.0-dev itself (run 29638976177) and independently affecting every currently-open PR built on top of it (fix(shutdown): stop wallet backend on GUI close and clear stale shutdown-channel log #905, fix: re-enable shielded transactions now that mainnet supports protocol v12 #906, feat(migration): auto-refresh DAPI nodes during pre-1.0 migration #908) — a shared/correlated CI-red issue, not caused by any of those PRs.
• What breaks without it: any PR based on v1.0-dev has a non-deterministic chance of a red "Tests" check on an otherwise-correct change, forcing needless re-runs and eroding trust in CI signal.
• Root cause: WalletBackend::stop_in_place() documents that pwm.spv().stop() releases the dash-spv storage advisory lock (LockFile, dash_spv::storage::lockfile) before returning, but the underlying OS-level flock's release isn't always synchronous with that under load — the test's immediate reconnect-in-place can lose a race against its own just-stopped predecessor. The upstream platform-wallet crate collapses the underlying error into an opaque PlatformWalletError::SpvError(String), so a structural (non-string-parsing) production-side fix isn't available from this repo without either a string-parsing hack (forbidden by this repo's conventions) or a change to the upstream crate (out of scope here).

What was done

• Replaced the single reconnect assertion in reconnect_restart_in_place_reuses_backend with a bounded, blind retry (up to 6 attempts, capped exponential backoff 25ms→400ms) around ensure_wallet_backend_and_start_spv. A genuine restart-in-place regression still fails loudly once the retry budget is exhausted — the retry does not inspect or match on error content (no string-parsing), it just retries any failure.
• Test-only change; no production code touched. Scope is limited to this one test — the other two backend_reopen_lock() users (issue7_fresh_persistor_bip44_xpub_matches_det_bridge, cold_boot_hydrates_persisted_transaction_history_without_live_events) exercise a different resource (the SqlitePersister single-open registry via a fresh AppContext) and never call .start() twice on the same backend instance, so they're unaffected by this race and were left untouched.
• Small doc-comment touch-up on the test to note the retry accommodates a timing window, not a change in the asserted production contract.

Testing

• Targeted test in a tight loop: 20/20 passed (initial), reconfirmed 15/15 after a wording edit.
• cargo test --all-features --workspace: full suite green.
• cargo clippy --all-features --all-targets -- -D warnings: clean.
• cargo +nightly fmt --all -- --check: clean.

Breaking changes

None — test-only change, no production behavior affected.

Checklist

• Tests pass locally
• Clippy clean
• Formatted
• No production code changed

Attribution

_{🤖 Co-authored by Claudius the Magnificent AI Agent}

Summary by CodeRabbit

• Bug Fixes
  • Improved wallet reconnection reliability after restarting the SPV service.
  • Reconnection now handles brief storage-lock timing delays automatically, reducing intermittent restart failures.

[coderabbitai]

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly summarizes the main change: hardening the reconnect-restart-in-place test against a dash-spv lock-release race.

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix/flaky-reconnect-in-place-test

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[thepastaclaw]

✅ Final review complete — no blockers (commit b36521f)

[thepastaclaw]

Final validation — Codex + Sonnet

Bounded, blind, capped-backoff retry (max 6 attempts) correctly absorbs a documented dash-spv storage-lock release timing race in a single offline test, without weakening the restart-in-place regression contract (still panics loudly on exhaustion). Verified against the exact head SHA; no in-scope issues found by either agent.

Review provenance

• Codex reviewers: gpt-5.6-sol — general (completed), gpt-5.6-sol — rust-quality (completed)
• Verifier: claude-sonnet-5 — final-verifier
• Sonnet reviewers: claude-sonnet-5 — general (completed), claude-sonnet-5 — rust-quality (completed)